Biometric-based self-service technology adoption by older adult: empirical evidence from pension fund sector in Indonesia

Abstract

Digital transformation brings new perspectives for companies in managing their businesses. One is about how companies accelerate information and service processes by utilizing technology. Due to security factors, biometric-based self-service technology (SST) is becoming popular and preferred in financial services. However, it is still relatively rare in developing countries like Indonesia, especially those specifically aimed at older adults. This study reveals the unique factors of older adults by investigating the effect of health conditions and facilitating conditions on the sustainable use of biometric-based self-service technology (SST) among older adult users. This research framework is based on the Theory of Planned Behavior and involved 298 elderly respondents in three provinces of Indonesia. The result shows that facilitating conditions and behavioral intentions to use are the main drivers of the continued use of biometric-based SST for elderly users. The declining health conditions of older adults (related to the ability to see, hear, and move limbs) affect their intention to use biometric-based self-service technology. Although health conditions were found not to influence the use behavior significantly, these factors still contribute to the sustainable use of biometric-based SST by elderly users. They are considered a concern in the study and development of future technologies.

Keywords:

• Self-service technology
• older adult
• biometrics
• self-authentication
• acceptance

Reviewing Editor:

• Jason Wang
• University of Huddersfield
• United Kingdom

Subjects:

• Consumer Psychology
• Strategic Management
• Management of Technology & Innovation

Introduction

The global population of people aged 65 and up is expected to more than double, rising from 761 million in 2021 to 1.6 billion in 2050 (United Nations, 2023). Data from the Central Bureau of Statistics of the Republic of Indonesia explain that from 1971 through 2019, the percentage of the elderly population in Indonesia increased by about two-fold to 9.6 percent (approx. 25 million people), and reached 10.28 percent in 2021 (Central Bureau of Statistics (Indonesia), 2021). Since 2021, Indonesia has entered an aging population structure, where around 1 in 10 of the population is elderly, and in 2022, the number of older adults increase to 10.48 percent (Central Bureau of Statistics (Indonesia), 2022). The older adults are a group that cannot be avoided by technological developments. Information and communication technology use in Indonesia has developed rapidly (Central Bureau of Statistics (Indonesia), 2019). Even though it is still relatively low, the technological literacy of older adults in Indonesia has increased from 5.73 percent in 2018 to 19.42 percent in 2022 (Central Bureau of Statistics (Indonesia), 2022). These data illustrate that older adults are currently a group with the potential to adopt technology, including service technology innovation.

During the COVID-19 pandemic, digital self-service has received much attention because social distance regulation necessitates non-contact services (Shi et al., 2023; Sixsmith et al., 2022). Self-service technology is now equipped with more sophisticated security methods like biometrics. Fingerprints, handprints, irises, voices, faces, signatures, ears, gaits, and other physiologic and behavioral traits of individuals are all examples of biometrics (Singh et al., 2021). Commercial biometrics deployments have been successful in financial, telecommunications, and other markets where authentication and fraud management are required (Kockmann et al., 2021). Several previous studies have analyzed the acceptance and use of biometric technology as security in consumer self-service, for example, banking services (Breward et al., 2017; Fatima, 2011; Lee & Pan, 2023), e-commerce (Al-Harby et al., 2010; Clodfelter, 2010; Hino, 2015), payment service (Moriuchi, 2021; Sulaiman & Almunawar, 2022; Zhong et al., 2021), and hospitality (Boo & Chua, 2022; Ciftci et al., 2021; Norfolk & O’Regan, 2020; Pai et al., 2018). However, securing only one biometric identification is not enough to improve customer security nowadays. Unimodal biometric systems cannot guarantee correct identification with certainty, whereas multimodal biometric systems, which employ multiple biometric modalities, ensure better recognition (Allam et al., 2022).

One of the pension fund services in Indonesia has implemented a biometric-based self-authentication service utilizing complex authentication and identification, including pin or password, face, eye, and voice, via a smartphone. The authentication of pension fund customers must be performed every month as proof that the pension beneficiaries are still alive. At the beginning of every month, retired civil servants will line up at the bank counters to be authenticated by the counter staff. For retirees who have reached old age and are beginning to become ill, their spouse or heirs can represent the disbursement of pension funds. Typically, the party representing the beneficiary must prepare a cover letter stating that the beneficiary is still alive and has the right to disburse the funds. A statement letter is needed to prevent disbursement of pension funds by individuals who are not entitled to receive them, or there is an element of fraud. There are cases where the retiree has died, but the heirs still take pension funds that are not their right (Kuswayati, 2019). The authentication process, previously limited to face-to-face interactions with officers at service counters, has recently been expanded to include self-authentication services enabled by a smartphone application on the customer’s device. This application’s authentication begins by entering an ID number and PIN or password. The application will then direct users to conduct a facial and eye biometric scan (accompanied by nodding, shaking the head, and blinking the eyes) and a voice for data matching. Upon successful authentication, pension beneficiaries can withdraw their pension funds from an automatic teller machine (ATM).

Biometric-based self-service is one way the pension fund company can expedite the service process while maintaining high security. However, the provision of biometric-based self-service technology in the form of smartphone applications for older adults is still relatively uncommon in Indonesia. Older people tend to use technology services less than younger groups due to factors such as digital literacy and access to the Internet (Anderson & Jiang, 2018; Hunsaker & Hargittai, 2018). Beyond these factors, personal and environmental contexts are also important to consider (Sixsmith et al., 2022). According to the Central Bureau of Statistics (Indonesia) (2022), slow access to technology by older adults in Indonesia is due to structural obstacles, namely the economic difficulties in accessing internet data packages or inadequate signals from internet service providers in the area where they live, and functional obstacles, which are characterized by a decline in the older adult health condition.

The older adult’s physical health is a crucial aspect of their continued use of technology since, as they age, their physiological and psychological abilities change, which has a significant impact on their demands and capacity to use technology or technical instruments (Farage et al., 2012; Tenneti et al., 2012). These modifications can occur against a person’s will. In several studies, physical condition is considered one of the individual factors that can predict technology adoption by older adults (e.g. Chen & Chan, 2014; Li et al., 2019; Mostaghel & Oghazi, 2017; Wang et al., 2017; Wang & Sun, 2016; Wilson et al., 2021). Age-related vision and hearing impairments might influence the usability of devices that rely primarily on graphical/text-based or sound-based user interfaces to deliver information (e.g. mobile phones and digital cameras) (Fisk et al., 2019; Wu et al., 2011). A study conducted by Wilson et al. (2021) found that one of the barriers for older people in using digital devices is due to the individual’s physical function, usually vision, for which they will choose a larger device, such as a tablet rather than a smartphone, because of the screen size. Older people may also have difficulty performing accurate movements (Cheong et al., 2013). For example, persistent pain, particularly in the fingers and wrists, was also a problem and impacted the way in which individuals used digital devices (Wilson et al., 2021). Even though it is not a technology that is used every day, self-authentication applications are run using smartphones, and biometric recognition requires the user to be precise in following system instructions to move body parts and match certain points so that authentication can be successful. These age-related differences in performance can be a major barrier to activities undertaken by older adults and must be considered for continued use of technology (Fisk et al., 2019).

In addition to the decline in older adults’ physical condition, which can impede the self-authentication process as they age, the availability of support can also determine the continued use of this technology. The biometric authentication application on a smartphone will scan certain points on a user’s body and match them with the biometric data that has been recorded. The supporting equipment (in this case, the smartphone’s camera and microphone) must be accurate and connected to a stable internet network. Self-service technology gives customers more flexibility and efficiency but requires skills and knowledge to use (Scherer et al., 2015). Customers participate actively in service delivery by interacting with platforms that provide opportunities for resource integration and encourage participation in product innovation or service improvement (Foroudi et al., 2018).

On the other hand, older adults lack confidence in their ability to use electronic devices to complete online tasks and require assistance in managing or utilizing these devices (Anderson & Perrin, 2017). In the customer technology acceptance and use context, perceptions of the availability of critical resources and assistance in carrying out specific activities are referred to as facilitating conditions (Venkatesh et al., 2012). Adequate facilitating conditions should play an essential role in the intention to use biometrics (Miltgen et al., 2013). For older adults, facilitating conditions are one of the variables having the greatest influence on the actual use of technology (Chen & Chan, 2014). In order to encourage older adults to use and become familiar with technology, incoming users of biometric-based SST should have more access to technical assistance and resources, such as user introductions and tutorials (Zhong et al., 2021).

Most of the studies on self-service technology, especially those using biometric technology, still involve young respondents or have not explicitly investigated the behavior of elderly users (e.g. Al-Harby et al., 2010; Norfolk & O’Regan, 2020; Soh et al., 2010; Zhong et al., 2021). Differences in technology adoption behavior between young adults and older people have been studied in several studies, either using age as a moderation (e.g. Kim & Bernhard, 2014; Venkatesh et al., 2003, 2012) or focusing on elderly users (e.g. Arenas-Gaitán et al., 2015; Chen & Chan, 2014; Ellis et al., 2021; Hoque & Sorwar, 2017; Hunsaker & Hargittai, 2018; Li et al., 2019; Mostaghel & Oghazi, 2017; Sixsmith et al., 2022; Tsai et al., 2020; Wang et al., 2017; Wilson et al., 2021). Studies focusing on elderly users have mostly been conducted in China and the United States, whereas those conducted in Southeast Asian countries are still very few (Yap et al., 2022). As one of the developing countries in Southeast Asia, Indonesia is also still lagging in its ability to adopt and explore digital technology to carry out transformations in government practices, business models, and society in general (ranked 51st out of 63 countries) (International Institute for Management Development, 2022). This research is important to encourage increased digital competitiveness in Indonesia. In addition, a study by Yap et al. (2022) shows most research focusing on elderly users is related to healthcare & assistive technology. Meanwhile, the relevance between online shopping and e-services is still minimal (Yap et al., 2022). In fact, along with technological developments, older adults are a group that has the potential to utilize increasingly diverse technology. This research aims to fill the gap by analyzing factors that can support or hinder older users’ continued use of biometric-based SST.

Previous studies have compared behavioral theories in the use of technology (e.g. Cheng, 2019; Ling et al., 2011; Rejali et al., 2023; Seok Kim et al., 2015). However, TPB still performs best in representing technology acceptance (Cheng, 2019; Rejali et al., 2023). TPB also remains valid in analyzing technology acceptance in elderly users (Choudrie & Vyas, 2014; Deng et al., 2014; Ellis et al., 2021). One of the main differences between TPB, the Technology Acceptance Model (TAM), and the Theory of Reasoned Action (TRA) is the treatment of behavioral control in TPB, referring to the skills, opportunities, and resources needed to use the system (Mathieson, 1991). According to TPB, success in carrying out a behavior plan depends not only on effort but also on the person’s control over other factors such as necessary information, skills, and abilities, including possession of a workable plan, will, presence of mind, time, opportunity, and so on (Ajzen, 1985).

This research adopts part of the TPB, namely the role of perceived behavioral control, to develop a model focusing on factors that support or hinder older adults from continuing to use biometric-based self-service technology. We use certain considerations to produce a more concise model. First, considering the results of a previous study conducted by Davis et al. (1989), Thompson et al. (1991), Taylor and Todd (1995), and Venkatesh et al. (2003) who stated that with the presence of other constructs in the model, attitude becomes insignificant towards behavioral intention. For this reason, we do not include attitude in our research model. Second, subjective norms refer to a person’s perception that the majority of people who are important to him think he should or should not carry out the behavior in question (Ajzen, 1991). In some studies, this is referred to as social influence. In several previous studies on biometric acceptance and technology use in older adults, social influence was found to have an insignificant effect on behavioral intention to use (e.g. Al-Harby et al., 2010; Arenas-Gaitán et al., 2015; Choudrie et al., 2020; Li et al., 2019; Miltgen et al., 2013; Sunandi & Koesrindartoto, 2019). Ellis et al. (2021) in their study found that family as a social influence source did not have a significant effect on the intention to use technology in older adults, both in the age group under 65 years and elderly aged 65 years and over. The reason for these results is that older adults often live far from family (children and/or grandchildren), so there is no advice from those closest to them about using technology (Ellis et al., 2021). However, another reason could be that older adults are more mature individuals and consider many things before behaving, including when using technology. The laggards group in the Diffusion of Innovation Theory (Rogers, 2003) is often associated with older adults (Lee & Coughlin, 2015; Peek et al., 2017), those who are very careful in adopting innovation. Thus, subjective norms or social influence are not really considered by older adults when using technology. Because this study aimed to analyze factors that may hinder and support the continued use of technology in older adults, we did not include subjective norms in the model.

Control beliefs in TPB can be individual (e.g. ability to use the system) and situational (e.g. having access to the system) (Mathieson, 1991). In older adults, age-related health conditions can be individualized supports or barriers to using self-authentication. For example, when eyesight declines, older adults may not be able to run a self-authentication application because they have difficulty seeing the small print on a smartphone and have difficulty adjusting the scanning area on the smartphone screen or because hearing loss means they have difficulty hearing the voice instructions of the application clearly, which can cause the wrong response. In other words, their reduced physical abilities prevent them from having the opportunity to use self-authentication applications. Facilitating conditions can be situational supports or barriers to using self-authentication. Elderly users who do not have resources, adequate knowledge about technology, and limited physical abilities due to age may experience difficulties, and using this technology will be hampered when they do not find someone who can help. In other words, they are not given the opportunity to use technology by situational factors.

Some control factors will be stable across situations, while others will vary from context to context (Ajzen, 1985). TPB utilizes important control variables for each situation independently and is more likely to capture those situation-specific factors (Mathieson, 1991). As suggested by Rejali et al. (2023) and Yap et al. (2022), the role of PBC in representing admissions requires further investigation. This research uses a modified TPB as a conceptual framework with more emphasis on perceived behavioral control (PBC) in the TPB. Perceived control over aspects related to older people’s technology use is important to trigger behavioral intentions (Yap et al., 2022).

Literature and hypothesis development

The theory of Planned Behavior by Ajzen (1985) is an extension of the Theory of Reasoned Action by Fishbein and Ajzen (1975). TPB added the construct of perceived behavioral control to the TRA model. Similar to TRA, TPB is predicated on the premise that humans often behave rationally, analyze available information, and implicitly or overtly weigh the consequences of their actions (Ajzen, 2005). Ajzen (1985) suggests that the development of TPB is also based on a number of assumptions, one of which is the ability of intention to foresee and explain that human conduct will be significantly affected when non-intentional circumstances exert a major influence on planned behavior. The theory of planned behavior examines the consequences of perceived behavioral control on the achievement of behavioral objectives. Intention primarily indicates a person’s willingness to attempt to engage in a specific conduct, whereas perceived control takes into account certain realistic limits.

Intentions can change throughout time, and any measure of intention gathered before the shift cannot be relied upon to predict future action reliably. Weak intention to do (or not perform) an activity result in less commitment, and relatively essential events can alter the intention or cause a change of mind. Some people are more prone to changing their minds than others (Ajzen, 1985). Hence, the association between intention and behavior is typically larger when the intention is held with high assurance. The accuracy of intention to behavior decreases when behavior is influenced by factors over which one has at least limited control.

Individual factor – health condition

Increasing age can result in physical challenges and health problems caused by decreased vision, hearing, and mobility/flexibility (Wood et al., 2010). There are limitations related to perception, cognition, and movement control that are increasing in prevalence (Fisk et al., 2019). The ability to use self-authentication in smartphone applications with complex biometric recognition can be impaired due to declining health conditions due to increasing age. In fact, periodic authentication (as proof of life) once a month is mandatory for pension benefit recipients in Indonesia and will be carried out until they die.

Visual impairment disproportionately affects older adults (Servat et al., 2011). Changes in the normal ability of the eye may occur in the 30s and will be more noticeable in the 40s (Kline & Scialfa, 1996). Vision changes also affect sensitivity to glare, breadth of the field, and processing speed (Erdinest et al., 2021; Fisk et al., 2019). The marked slowdown of visual information processing increases with age, gradually reducing the perceptual flexibility of visual sensations (Fisk et al., 2019). Like vision, hearing loss is also closely related to age. Older people generally experience moderate hearing loss (Kline & Scialfa, 1996). Hearing loss affects one-third of adults aged 60–70 and two-thirds of adults by the time they reach 70; every ten years, the frequency of older adults with hearing loss doubles (Contrera et al., 2016). Approximately 10 percent of all middle-aged adults have hearing loss, and by age 65 and over, the percentage jumps to more than half of all men and 30 percent of all women (Fisk et al., 2019). Reduced hearing can result in mishearing and responding illogically (Hyams et al., 2018).

In the previous study, age-related health that affects the physical was constructed as perceived physical condition (Deng et al., 2014), biophysical aging restrictions (Wang et al., 2017) and health condition (Chen & Chan, 2014; Ha & Park, 2020; Li et al., 2019; Mostaghel & Oghazi, 2017). Chen and Chan (2014) measure health conditions with respect to five items: general health status, health conditions compared with the same-age groups, visual ability, auditory ability and movement ability. Meanwhile, Li et al. (2019) measured health conditions with three items: health status, health status compared with peers, and the ability of auditory, visual and mobility. In this study, health condition refers to an individual’s perception/view of their physical health condition that allows them to use self-authentication technology. According to Wang et al. (2017), physical aging, which includes visual acuity, hearing, and dexterity, must be considered in predicting technology acceptance in older adults because the technology industry must carry out market positioning and consider the demands and physiological conditions of older adults as the starting point for product development.

Li et al. (2019), in a study about health smart clothing technology, found that health conditions affect behavioral intention in using technology. Health conditions were also found to influence behavioral intention to use in research on digital games (Wang & Sun, 2016) but not significantly in studies on cell phones (Wang et al., 2017). Health conditions were found to directly predict technology usage (Chen & Chan, 2014; Ha & Park, 2020). The use of technology for older adults needs further research on different technologies to fill in the gaps in the results of previous studies (Yap et al., 2022). The user’s ability is essential because health conditions tend to decline with age, which can be a barrier to using self-authentication applications. Older adult may not be able to run self-authentication applications because they have difficulty seeing small print on a smartphone and are less precise when adjusting the scanning area on the phone screen. Hearing loss may make it difficult for them to hear the app’s voice instructions clearly. Therefore, it is hypothesized that:

H₁: Health conditions significantly affect the intention to use self-authentication applications.

H₂: Health conditions significantly affect the use behavior of self-authentication applications.

Situational factor – facilitating condition

TPB suggests that individual external factors determining the occurrence of behavior are perceived facilities. Perceived facilities in TPB (facilitating conditions) refer to individual judgments about the importance of resources for achieving results (Mathieson, 1991). Facilitating conditions may also involve motivation or environmental obstacles that influence an individual’s perception of the ease or difficulty of carrying out particular actions (Taylor & Todd, 1995). Venkatesh et al. (2003) conceptualized facilitating conditions from three constructs: perceived behavioral control (TPB/Decomposed TPB, Combined TAM-TPB), facilitating conditions (Model of PC Utilization), and compatibility (Innovation Diffusion Theory), which were operationalized to include aspects of the technological and organizational environment that designed to remove barriers to use. Facilitation conditions are defined as the extent to which a person believes that there is an organizational and technical infrastructure to support the use of the system (Venkatesh et al., 2003). In the context of the customer, facilitating conditions can be characterized as the customer’s perception of the availability of essential resources and help for carrying out particular tasks (Venkatesh et al., 2012). Related to age, the effect of facilitating conditions on behavioral intention is more pronounced in older women as this consumer group considers the availability of resources, knowledge, and support to be necessary for the continued use of technology (Venkatesh et al., 2012).

The research results of Chen and Chan (2014) show the importance of facilitating conditions (knowledge, guidance and support from other people, and a level of accessibility) that allow the older population to use gerontechnology. Choudrie et al. (2020), in their research about older adults’ adoption of mobile phones, found that facilitating condition (including resources, knowledge, operating cost and the person available for assistance) influences behavioral intention to use. However, facilitation conditions were found to have no significant effect on behavioral intention to use mHealth (Hoque & Sorwar, 2017). The effect of facilitating conditions in previous studies involving elderly users has not shown consistent results (Yap et al., 2022).

Facilitating conditions are an important factor that influences behavioral intentions to use biometric technology (Miltgen et al., 2013; Ngugi et al., 2011; Zhong et al., 2021). However, inconsistent results were still found. A study by Zhong et al. (2021) in South Korea found that facilitating conditions were a vital variable that was statistically significant to behavioral intention to use facial recognition payment. However, research by Ciftci et al. (2021) in the United States found that facilitating conditions did not have a significant effect on behavioral intention to use facial recognition systems in quick-service restaurants. In the current research, the Self-Authentication application used by pension benefit recipients uses multimodal biometric systems. In order to carry out online self-authentication, the user’s smartphone must be connected to the internet network. Apart from that, compatible technological support is needed because in operation this system will scan certain points on an individual’s body, including voice and match it with biometric data that has been recorded. Of course, the supporting equipment used, for example the camera and mic on a smartphone, must be has good accuracy. In addition, due to limited physical abilities caused by increasing age, they will really need the help of other people when they experience problems in using this technology. Facilitating conditions can be important for technology users heavily relying on technical support and other resources to adopt the technology. This will affect the behavioral intention and the continuance use self-authentication application. Therefore, it is hypothesised that:

H₃: Facilitating conditions significantly affect the intention to use self-authentication applications.

H₄: Facilitating conditions significantly affect the use behavior of self-authentication applications.

Behavioral intention to use and use behavior

TPB is based on the assumption that humans usually behave reasonably, consider available information, and implicitly or explicitly consider the implications of their actions (Ajzen, 2005). Behavior is determined by the intention to do so. Nonetheless, it was found in a study on biometric technology that behavioral intention to use the technology did not have a significant effect on use behavior (Kanak & Sogukpinar, 2017). However, another study on smartphone adoption, use and diffusion among older adults in the UK found that behavioral intention to use influences use behavior (Choudrie et al., 2020). Therefore, it is hypothesized that:

H₄: Behavioral intention to use significantly affects the use behavior of self-authentication applications.

H₅: Behavioral intention to use significantly affects the use behavior of self-authentication applications.

The research framework of this study is depicted in Figure 1.

Figure 1. Purposed research framework.

Methodology

Measurement development

This study examines the research model, the influence of health conditions and facilitating conditions on behavioral intention in using technology, and how they impact use behavior. This study used a quantitative survey method to test the proposed research model empirically. Questionnaire items were developed from previous research (Table 1.), including statements about self-reported health conditions, facilitating conditions, behavioral intention to use, and use behavior. Health condition items include general health status, health conditions compared to the same age group, visual ability, hearing ability, and movement ability (Chen & Chan, 2014).

Table 1. Measurement item of construct.

Items	Author
Health condition HC1: I have a good health condition when compared to people my age. HC2: I have good vision conditions to use apps on smartphones. HC3: I can hear sounds/voices clearly. HC4: I can move my limbs well to perform biometric authentication with applications.	Chen and Chan (2014)
Facilitating condition FC1: My financial ability helps me overcome difficulties in accessing the self-authentication application. FC2: My knowledge of technology helps me overcome difficulties in using the app. FC3: The instructions provided in the app can help me overcome difficulties in using the application. FC4: I have a sufficient mobile device to use the app. FC5: Family/people around help overcome difficulties in using the app.	Chen and Chan (2014); Zhong et al. (2021)
Behavioral intention to use BI1: I intend to use the application again in the future. BI2: I will use the application independently whenever I take a pension fund. BI3: I will continue to use the app for authentication rather than to the counter (bank). BI4: I would suggest other retirees use the self-authentication application.	Morosan (2012); Zhong et al., 2021)
Use behavior UB1: Using the self-authentication application is a pleasant experience. UB2: I did self-authentication with the authentication application on my smartphone. UB3: Before withdrawing my pension fund, I always authenticate myself with the self-authentication application on my smartphone without coming to the bank/post office.	Hoque and Sorwar (2017)

Facilitating condition items include financial resources, basic knowledge, individual or group assistance/support, technology support (compatible), and instructions (Chen & Chan, 2014; Zhong et al., 2021). Behavioral intention to use items consists of intention to use, willingness to use, plans to continue using the technology despite alternatives to counter services, and intentions to suggest others to use self-authentication applications (Morosan, 2012; Zhong et al., 2021). Use behavior items include using self-authentication applications as a pleasurable experience, self-reported use, and frequency of use (Hoque & Sorwar, 2017).

Variables were measured using a Likert scale of 5, in which ‘1 = strongly disagree’ to ‘5 = strongly agree’. The questionnaire was designed in two parts; the first part captured information regarding the respondent’s demographic profile, and the second part captured information regarding the user’s assessment of their health conditions, facilitating conditions, behavioral intention and behavior in using biometric-based SST. The research structural model is shown in Figure 2.

Figure 2. Structural model.

Sample and data collection

Since the self-authentication application was implemented, this application has not been used by all retired civil servants. As a result of the study’s purpose, the target population in this study is retired civil servants (age 60 and above) who have used self-authentication applications. The G*Power program calculates the number of samples in this study with the following parameters: statistical power of 80 percent, significant level of 5 percent, and minimum R2 of 0.20. The 80 percent determination of statistical power is based on the opinion of Hair et al. (2017) that the level of commonly used statistical power is 80 percent. The minimum R² is determined at 0.20 since an R² value of 0.20 is considered high in disciplines such as consumer behavior (Hair et al., 2017). Thus, the minimum sample in this study was 193 respondents.

A total of 360 offline questionnaires were distributed in three provinces in Indonesia, namely East Java, Central Java, and West Java. These three provinces have the highest number of retired civil servants in Indonesia. The questionnaire was distributed in collaboration with the Association of Wredatama Republik Indonesia, a social organization in Indonesia whose members are retired civil servants. The questionnaire given to prospective respondents was accompanied by a statement from the researcher to ensure the confidentiality of the participant’s data. The answers will be used only for research purposes. The questionnaire was then filled out by individuals who had agreed and were willing to be respondents. All respondents met the criteria for being elderly according to the provisions in Indonesia, namely, aged 60 years and over. There was a total of 298 valid responses for analysis. According to Hair et al. (2017), the ideal sample size for structural equation modeling is between 100 and 200. This study met the criterion with a sample size of more than 200. The demographic profile of the respondents in Table 2 illustrates that 47 percent of men and 53 percent of women participated in this study. Most respondents (56.0%) are between 60 and 69 years old, and the majority (49.0%) have Bachelor’s degrees.

Table 2. Sample profile.

Demographic Variables		Frequencies	Percent (%)
Age	60–69	167	56.0
	70–79	121	40.6
	Above 80	10	3.4
Gender	Male	162	54.4
	Female	136	45.6
Education	Below diploma	44	14.8
	Diploma	57	19.1
	Degree	146	49.0
	Master	38	12.8
	Doctor/PhD	13	4.4

The proposed model was evaluated using the PLS-SEM (partial least squares structural equation modeling) technique with WarpPLS software version 7. In examining complex interactions between numerous variables, PLS is often useful for avoiding invalid solutions and factor indeterminacies (Fornell & Bookstein, 1982). This approach has been widely used in several business studies in recent years (He et al., 2020; Xu et al., 2020). In PLS analysis, the measurement model is frequently referred to as the outer model, while the structural model is referred to as the inner model (Hair et al., 2018).

Result

Measurement model

The measuring model is evaluated based on its internal consistency, convergent validity, and discriminant validity (Hair et al., 2017). In accordance with Hair et al. (2017), AVE and CR values over 0.5 and 0.70, respectively, imply that the model has the same construct dependability. Tables 3 and 4 exhibit the loading factor, AVE, composite reliability, and Cronbach’s alpha (α). In Table 3, the estimated Cronbach’s alpha (α) values vary from 0.852 to 0.885, while the composite reliability values range from 0.891 to 0.921, indicating a high degree of internal consistency. Table 3 further reveals that the loading factor values vary from 0.670 to 0.918, and the AVE values range from 0.624 to 0.785, which is over the recommended limit. Thus, the criteria of convergent validity are satisfied in the present investigation. The square root of the AVE and the cross-loading matrix were used to evaluate discriminant validity. For discriminant validity, the square root of the AVE of a construct must be greater than its correlation with other constructs (Hair et al., 2017). Table 4. demonstrates that all constructs from this investigation support the discriminant validity of the data.

Table 3. Measurement model assessment.

Item	Loading factor	AVE	CR	Cronbach’s alpha
Health condition
HC1	0.813	0.700	0.903	0.857
HC2	0.818
HC3	0.866
HC4	0.848
Facilitating condition
FC1	0.869	0.688	0.916	0.885
FC2	0.845
FC3	0.900
FC4	0.786
FC5	0.736
Behavioral intention to use
BI1	0.954	0.877	0.966	0.953
BI2	0.929
BI3	0.925
BI4	0.936
Use behavior
UB1	0.940	0.884	0.958	0.934
UB2	0.959
UB3	0.921

Table 4. Latent variable correlations and square-roots of aves.

	HC	FC	BI	UB
HC	(0.837)	0.544	0.479	0.539
FC	0.544	(0.829)	0.741	0.754
BI	0.479	0.741	(0.936)	0.917
UB	0.539	0.754	0.917	(0.940)

The bold value emphasizes that this value is higher than the correlation value of other constructs.

Structural model

The structural model was developed to identify the relationship between constructs in the research model. This study examines the relationship between the dependent and independent variables with a path coefficient (β) and a p-value <0.05. Based on the results of structural model testing (Figure 3. and Table 5.), health conditions (HC) have a significant effect on behavioral intention to use (BI) (β = 0.164; p < 0.05), which confirms H1, but has no significant effect on Use behavior (UB) (β = 0.040; p > 0.05), which rejected H2. Behavioral intention to use (BI) is significantly influenced by facilitating condition (FC) (β = 0.552; p < 0.05), which confirms H3. Use behavior (UB) is significantly influenced by facilitating condition (FC) (β = 0.265; p < 0.05) and behavioral intention to use (BI) (β = 0.621; p < 0.05), which confirms H4 and H5.

Figure 3. Structural model measurement.

Table 5. Path coefficients.

Hypothesis		β	SE	p-Value	Result
H1	HC → BI	0.164	0.089	0.000***	Accepted
H2	HC → UB	0.040	0.051	0.122	Not accepted
H3	FC → BI	0.552	0.071	0.000***	Accepted
H4	FC → UB	0.123	0.065	0.000***	Accepted
H5	BI → UB	0.779	0.047	0.000***	Accepted

The model explained 65.6 percent of the variance in use behavior (R² = 0.66) and 35 percent in intention to use (R² = 0.35). In addition to the R-squared, the Q-squared (Q²) was used to assess the predictive relevance of the model (Hair et al., 2017). The models had predictive relevance because the Q² was greater than zero (Q² = 0.779).

Discussion

This study aimed to determine the effect of health conditions and facilitating conditions on intention to use and their impact on the continued use of biometric-based self-service technology among elderly users in a developing country. The results revealed that health conditions and facilitating conditions significantly influence behavioral intention to use, which is consistent with TPB (Fishbein & Ajzen, 1975). Based on TPB, the ability of intention to predict and explain human behavior will be significantly influenced by non-intentional factors. Our findings show that a positive effect of declining health conditions will tend to reduce the intention of elderly users to use self-authentication applications or, in other words, a decrease in older adult health conditions, including the ability to see, hear, the ability to move limbs will affect their behavioral intention to use self-authentication application. Of course, this can happen, considering that authentication with self-service technology based on biometrics is carried out using a cellular phone. The users must make movements according to specific instructions to match authentication points such as eyes and head movements. Obstacles that arise from a decline in physical condition due to age can make older people ultimately less interested in using self-authentication applications. Their reduced physical abilities prevent them from having the opportunity to use self-authentication applications. This self-service seems more difficult, and counter service may be the best option.

The age of the respondents strengthens this result. The majority of respondents in this study were 60–69 years, or 56 percent. Still, the other 44 percent were aged 70 years and over, where the decline in physical condition in older people aged 70 years and over is often more significant than in those in their 60s. The reduction in vision becomes more pronounced and hearing also decreases further (Contrera et al., 2016; Fisk et al., 2019; Kline & Scialfa, 1996). In addition, obstacles in movement and mobility become more severe. This result aligns with the previous study by Wang and Sun (2016) and Li et al. (2019). In line with the opinion of Wang et al. (2017) that physical aging affecting visual acuity, hearing, and dexterity must be considered in predicting technology acceptance for older adults because the technology industry must carry out market positioning and consider the demands and physiological conditions of older adults as a product development starting point.

Facilitating conditions affect behavioral intention to use and use behavior directly. This effect confirms that facilitating conditions such as adequate resources and support are very important for elderly users to encourage their intention to use technology sustainably. This result aligns with previous research stated that facilitating conditions must play an important role in the biometrics technology adoption (i.e. Miltgen et al., 2013; Ngugi et al., 2011; Zhong et al., 2021), but different from the results of previous studies by Ciftci et al. (2021) that showed an insignificant effect of this construct on the intention to use facial recognition system. In developed countries such as the United States, consumers may perceive face authentication as a technology that does not require additional training, guidance, or support (i.e. facilitating conditions) to use it because this technology is implemented in many different settings for personal use, including unlocking personal devices and identity verification in computer games and social media (Ciftci et al., 2021). These previous studies did not focus on elderly consumers. Therefore, this research adds evidence to the knowledge that the continued use of biometric-based SST by elderly consumers in developing countries is highly dependent on situational factors: financial ability, knowledge of technology tutorial support, sufficient mobile device technology, and help from other people.

In Indonesia, internet access, on average, is still paid. For this reason, respondents in this study consider financial ability to be important for using self-authentication applications. Basic knowledge of technology can help seniors overcome difficulties in using self-authentication applications. Adequate basic knowledge is reinforced by the description of respondents, which shows that the majority of respondents in this study had a bachelor’s degree (49.0%). Riddell and Song (2017) state that employees with higher education tend to use computers at work. Graduating from high school increases the probability of using a computer at work by 37 percent, and an additional year of school increases that probability by 7 percent; these figures have a large impact in size and statistically (Riddell & Song, 2017). The biometric authentication system is slightly different in its use because it is related to the support of compatible technology (in this case, the camera and mic on a smartphone), which must have good accuracy to scan certain points on an individual’s body and match it with the biometric data that has been recorded. The decline in the ability of older people due to increasing age makes them dependent on other people to guide them in using self-authentication applications, but SST distances customers from service personnel, therefore, instructions available in the application and support from people around them become important. Older adult users may eventually be unable to use this technology if no one helps them. Facilitating conditions are one of the variables having the most significant influence on the older adults actual use of technology (Chen & Chan, 2014).

The interesting thing about this study is that the effect of health conditions on use behavior is insignificant. It is inconsistent with TPB, which emphasizes that success in carrying out a behavior plan (actual behavior) depends not only on effort but also on the person’s control over other factors, one of which is ability (Ajzen, 1985). This result also differs from previous research stating that health conditions directly affect the actual use of technology (Chen & Chan, 2014). Although not significant, the effect of health conditions on use behavior shows a positive direction. Our research finding contradicts the study by Chen and Chan (2014), which shows that health conditions negatively affect the usage behavior of gerontechnology, or it can be said that people with poor health conditions are more likely to use technology. The difference could be due to the different types of technology examined in our study. The insignificant effect of health conditions on use behavior means that declining health conditions does not sufficiently influence the use behavior of self-authentication application. This result is possible because of the availability of supporting conditions that can assist older adults in using technology even though they have limited abilities due to age. Based on the average respondent’s statement on the facilitating condition variable, it was found that the items regarding the availability of support from family members/people around them has the highest score. It means that the assistance from family members/people around plays an important role in supporting elderly users to continue using biometric-based self-service technology. This result is supported by statistical data on the elderly population of Indonesia 2021 (Central Bureau of Statistics (Indonesia), 2021) which shows that more older people in Indonesia (34.71 percent) live with three generations in a household. Living with three generations means that the elderly lives with their children and grandchildren in one house, or with their children and parents. Furthermore, 29.66 percent of the elderly live with their nuclear family and 22.78 percent live with their partners.

Behavioral intention to use has the most significant influence on use behavior (0.779) compared to facilitating conditions (0.123). Our findings support the TPB (Ajzen, 1985), indicating that the individual’s intention to conduct a behavior is the primary factor in the Theory of Planned Behavior. It is considered that intention captures the motivational variables that influence action; intention reveals how hard people are willing to try and how much effort they intend to exert to achieve a behavior. Intention, in turn, is seen as one of the direct antecedents of actual behavior. The stronger people’s intentions to engage in a behavior or achieve their behavioral goals, the more predictably successful they will be. The higher the behavioral intention of elderly users to authenticate through biometric-based self-service technology, the higher the actual use or behavior of using biometric-based self-service technology will be. Our result aligns with the research of Demoulin and Djelassi (2016), which found that behavioral intention to use self-service technology positively influences use behavior, and research of Choudrie et al. (2020) on smartphones among older adults in England, which found that behavioral intention influences use behavior.

Conclusion

Reduced health condition (general health condition, visual acuity, hearing, and movement) is a barrier for older adults to continue using self-authentication applications. In order to maintain the continued use of biometric-based SST, it is necessary to pay special attention to the unique characteristics of older adults, particularly in services designed for older people. On the other hand, older adults depend heavily on facilitating as a situational factor (technical support and resources) that can encourage them to continue using self-service technology.

This study summarizes the TPB model by focusing on the importance of perceived behavioral control aspects, developing and validating a model to identify the unique characteristics of older adults that may hinder and support the continued use of biometric-based self-service technologies by older adults in developing countries. The results of the model testing demonstrated high predictive (Q²=0.78) and explanatory power (R²=0.66) for all dependent variables. This study provides additional support for the use of TPB in the implementation of biometric self-service technology.

The findings of this research provide practical implications for the benefits of increasing the application of self-service technology for older adults in developing countries, especially in Indonesia, one of the developing countries currently experiencing a demographic bonus. This research contributes to the government’s increasing technology adoption in all age groups and the creating of an elderly-friendly digital ecosystem. According to Laudon and Laudon (2016), behavioral problems are an important part that arises in the development and long-term maintenance of information systems. The behavioral approach focuses not on technical solutions but on changing attitudes, management and organizational policies, and behavior (Laudon & Laudon, 2016). In order to increase service efficiency and achieve long-term success, businesses must improve their services with the latest technology while remaining focused on elements that encourage sustainable use of service technology. Companies as service providers need to consider the physical condition of older adults and the availability of accessible support to develop service technology based on customer needs and avoid wasted investment due to discontinuation of technology use.

This study presents theoretical and practical insights that can be used to predict the adoption of biometric self-service technologies for older adults in other developing countries. However, there are several limitations in this research. First, respondents were limited to a certain age, namely 60 years and over, who met the criteria for older people in Indonesia. The age limit for older people in each country can vary. Second, a specific location is selected to carry out the survey. This research was conducted in 3 provinces in Indonesia, and the technology literacy index was different in each province. Therefore, further research could cover a wider geographical area or in different countries. Third, the health conditions in this study are health conditions self-reported by respondents. There are two ways to measure health condition variables, namely (1) using a person’s health condition based on medical reports and (2) using self-reported health. Self-reported depends on the accuracy of the respondent’s assessment. This method was used because it is easier to process and cheaper because it does not require special equipment (Tenneti et al., 2012). Future research can use measurements of health conditions in various ways, namely by measuring a person’s health condition or based on medical reports. Finally, this research focuses on health conditions and facilitating conditions as control variables in the TPB. Future research can expand this research model by involving other variables.

Acknowledgment

The research received no research funding.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The datasets used and/or analyzed during this investigation are available upon reasonable request from the corresponding author.

Additional information

Notes on contributors

R. Amalina Dewi Kumalasari

R. Amalina Dewi Kumalasari is a Doctoral program student at the Faculty of Administration Science, Brawijaya University, Indonesia. She holds a Bachelor’s and Master’s in Business Administration from the Faculty of Administration Science the University of Brawijaya. Her main focus of research is on technology user behavior.

Kusdi Rahardjo

Kusdi Raharjo is a Professor in the Business Administration Department, Faculty of Administrative Science, Brawijaya University, Indonesia. He obtained his doctorate in Gestion des Organization from the Institute d’Admnistration des Entreprises, Nice Sophia Antipolis University, France. His research focuses on human resources. He actively participates in seminars domestically and abroad as a speaker and participant.

Andriani Kusumawati

Andriani Kusumawati is an Associate Professor in the Business Administration Department, Faculty of Administrative Science, Brawijaya University, Indonesia. She obtained his doctorate from Sydney Business School, the University of Wollongong, Australia, in Marketing Management with a DBA degree. Her research focuses on marketing, consumer behavior, service marketing, tourism marketing, and education marketing.

Sunarti Sunarti

Sunarti is an Assistant Professor in the Business Administration Department, Faculty of Administrative Science, Brawijaya University, Indonesia. She obtained his doctorate from the University of Airlangga, Indonesia. Her research focuses on marketing.